Acoustic resonators, such as Surface Acoustic Wave (SAW) resonators and Bulk Acoustic Wave (BAW) resonators, are used in many high-frequency communication applications. In particular, SAW resonators are often employed in filter networks that operate frequencies up to 1.8 GHz, and BAW resonators are often employed in filter networks that operate at frequencies above 1.5 GHz. Such filters need to have flat passbands, have steep filter skirts and squared shoulders at the upper and lower ends of the passband, and provide excellent rejection outside of the passband. SAW- and BAW-based filters also have relatively low insertion loss, tend to decrease in size as the frequency of operation increases, and are relatively stable over wide temperature ranges. As such, SAW- and BAW-based filters are the filter of choice for many 3rd Generation (3G) and 4th Generation (4G) wireless devices and are destined to dominate filter applications for 5th Generation (5G) wireless devices. Most of these wireless devices support cellular, wireless fidelity (Wi-Fi), Bluetooth, and/or near field communications on the same wireless device and, as such, pose extremely challenging filtering demands. While these demands keep raising the complexity of wireless devices, there is a constant need to improve the performance of acoustic resonators and filters that are based thereon.
To better understand acoustic resonators and various terminology associated therewith, the following provides an overview of a BAW resonator. However, the concepts described herein may employ any type of acoustic resonator and are not limited to SAW- and BAW-based resonators. An exemplary BAW resonator 10 is illustrated in FIG. 1. The BAW resonator 10 generally includes a substrate 12, a reflector 14 mounted over the substrate 12, and a transducer 16 mounted over the reflector 14. The transducer 16 rests on the reflector 14 and includes a piezoelectric layer 18, which is sandwiched between a top electrode 20 and a bottom electrode 22. The top and bottom electrodes 20 and 22 may be formed of Tungsten (W), Molybdenum (Mo), Platinum (Pt), or like material, and the piezoelectric layer 18 may be formed of Aluminum Nitride (AlN), Zinc Oxide (ZnO), or other appropriate piezoelectric material. Although shown in FIG. 1 as each including a single layer, the piezoelectric layer 18, the top electrode 20, and/or the bottom electrode 22 may include multiple layers of the same material, multiple layers in which at least two layers are different materials, or multiple layers in which each layer is a different material.
The BAW resonator 10 is divided into an active region 24 and an outside region 26. The active region 24 generally corresponds to the section of the BAW resonator 10 where the top and bottom electrodes 20 and 22 overlap and also includes the layers below the overlapping top and bottom electrodes 20 and 22. The outside region 26 corresponds to the section of the BAW resonator 10 that surrounds the active region 24.
For the BAW resonator 10, applying electrical signals across the top electrode 20 and the bottom electrode 22 excites acoustic waves in the piezoelectric layer 18. These acoustic waves primarily propagate vertically. A primary goal in BAW resonator design is to confine these vertically propagating acoustic waves in the transducer 16. Acoustic waves traveling upward are reflected back into the transducer 16 by the air-metal boundary at the top surface of the top electrode 20. Acoustic waves traveling downward are reflected back into the transducer 16 by the reflector 14 or by an air cavity, which is provided just below the transducer in a Film BAW Resonator (FBAR).
The reflector 14 is typically formed by a stack of reflector layers (RL) 28, which alternate in material composition to produce a significant reflection coefficient at the junction of adjacent reflector layers 28. Typically, the reflector layers 28 alternate between materials having high and low acoustic impedances, such as tungsten (W) and silicon dioxide (SiO2). While only five reflector layers 28 are illustrated in FIG. 1, the number of reflector layers 28 and the structure of the reflector 14 varies from one design to another.
The magnitude (Z) and phase (ϕ) of the electrical impedance as a function of the frequency for a relatively ideal BAW resonator 10 is provided in FIG. 2. The magnitude (Z) of the electrical impedance is illustrated by the solid line, whereas the phase (ϕ) of the electrical impedance is illustrated by the dashed line. A unique feature of the BAW resonator 10 is that it has both a resonance frequency and an anti-resonance frequency. The resonance frequency is typically referred to as the series resonance frequency (fs), the anti-resonance frequency is typically referred to as the parallel resonance frequency (fp). The series resonance frequency (fs) occurs when the magnitude of the impedance, or reactance, of the BAW resonator 10 approaches zero. The parallel resonance frequency (fp) occurs when the magnitude of the impedance, or reactance, of the BAW resonator 10 peaks at a significantly high level. In general, the series resonance frequency (fs) is a function of the thickness of the piezoelectric layer 18 and the mass of the bottom and top electrodes 20 and 22.
For the phase, the BAW resonator 10 acts like an inductance that provides a 90° phase shift between the series resonance frequency (fs) and the parallel resonance frequency (fp). In contrast, the BAW resonator 10 acts like a capacitance that provides a −90° phase shift below the series resonance frequency (fs) and above the parallel resonance frequency (fp). The BAW resonator 10 presents a very low, near zero, resistance at the series resonance frequency (fs) and a very high resistance at the parallel resonance frequency (fp). The electrical nature of the BAW resonator 10 lends itself to the realization of a very high Q (quality factor) inductance over a relatively short range of frequencies, which has proved to be very beneficial in high-frequency filter networks, especially those operating at frequencies around 1.8 GHz and above.
Unfortunately, the phase (ϕ) curve of FIG. 2 is representative of an ideal phase curve. In reality, approaching this ideal is challenging. A typical phase curve for the BAW resonator 10 of FIG. 1 is illustrated in FIG. 3A. Instead of being a smooth curve, the phase curve of FIG. 3A includes ripple below the series resonance frequency (fs), between the series resonance frequency (fs) and the parallel resonance frequency (fp), and above the parallel resonance frequency (fp). The ripple is the result of spurious modes, which are caused by spurious resonances that occur in corresponding frequencies. While the vast majority of the acoustic waves in the BAW resonator 10 propagate vertically, various boundary conditions about the transducer 16 result in the propagation of lateral (horizontal) acoustic waves, which are referred to as lateral standing waves. The presence of these lateral standing waves reduces the potential Q associated with the BAW resonator 10.
As illustrated in FIG. 4, a border (BO) ring 30 is formed on or within the top electrode 20 to suppress certain of the spurious modes. The spurious modes that are suppressed by the BO ring 30 are those above the series resonance frequency (fs), as highlighted by circles A and B in the phase curve of FIG. 3B. Circle A shows a suppression of the ripple, and thus of the spurious mode, in the passband of the phase curve, which resides between the series resonance frequency (fs) and the parallel resonance frequency (fp). Circle B shows suppression of the ripple, and thus of the spurious modes, above the parallel resonance frequency (fp). Notably, the spurious mode in the upper shoulder of the passband, which is just below the parallel resonance frequency fp, and the spurious modes above the passband are suppressed, as evidenced by the smooth or substantially ripple free phase curve between the series resonance frequency (fs) and the parallel resonance frequency (fp) and above the parallel resonance frequency (fp).
The BO ring 30 corresponds to a mass loading of the portion of the top electrode 20 that extends about the periphery of the active region 24. The BO ring 30 may correspond to a thickened portion of the top electrode 20 or the application of additional layers of an appropriate material over the top electrode 20. The portion of the BAW resonator 10 that includes and resides below the BO ring 30 is referred to as a BO region 32. Accordingly, the BO region 32 corresponds to an outer, perimeter portion of the active region 24 and resides inside of the active region 24.
While the BO ring 30 is effective at suppressing spurious modes above the series resonance frequency (fs), the BO ring 30 has little or no impact on those spurious modes below the series resonance frequency (fs), as shown by the ripples in the phase curve below the series resonance frequency (fs) in FIG. 3B. A technique referred to as apodization is often used to suppress the spurious modes that fall below the series resonance frequency (fs).
Apodization tries to avoid, or at least significantly reduce, any lateral symmetry in the BAW resonator 10, or at least in the transducer 16 thereof. The lateral symmetry corresponds to the footprint of the transducer 16, and avoiding the lateral symmetry corresponds to avoiding symmetry associated with the sides of the footprint. For example, one may choose a footprint that corresponds to a pentagon instead of a square or rectangle. Avoiding symmetry helps reduce the presence of lateral standing waves in the transducer 16. Circle C of FIG. 3C illustrates the effect of apodization in which the spurious modes below the series resonance frequency (fs) are suppressed, as evidence by the smooth or substantially ripple free phase curve below the series resonance frequency (fs). Assuming no BO ring 30 is provided, one can readily see in FIG. 3C that apodization fails to suppress those spurious modes above the series resonance frequency (fs). As such, the typical BAW resonator 10 employs both apodization and the BO ring 30.
As noted previously, BAW resonators 10 are often used in filter networks that operate at high frequencies and require high Q values. A basic ladder network 40 is illustrated in FIG. 5A. The ladder network 40 includes two series resonators BSER and two shunt resonators BSH, which are arranged in a traditional ladder configuration. Typically, the series resonators BSER have the same or similar first frequency response, and the shunt resonators BSH have the same or similar second frequency response, which is different from the first frequency response, as shown in FIG. 5B. In many applications, the shunt resonators BSH are detuned versions of the series resonators BSER. As a result, the frequency responses for the series resonators BSER and the shunt resonators BSH are generally very similar, yet shifted relative to one another such that the parallel resonance frequency (fp,SH) of the shunt resonators approximates the series resonance frequency (fs,SER) of the series resonators BSER. Note that the series resonance frequency (fs,SH) of the shunt resonators BSH is less than the series resonance frequency (fs,SER) of the series resonators BSER. The parallel resonance frequency (fp,SH) of the shunt resonators BSH is less than the parallel resonance frequency (fp,SER) of the series resonators BSER.
FIG. 5C is associated with FIG. 5B and illustrates the response of the ladder network 40. The series resonance frequency (fs,SH) of the shunt resonators BSH corresponds to the low side of the passband's skirt (phase 2), and the parallel resonance frequency (fp,SER) of the series resonators BSER corresponds to the high side of the passband's skirt (phase 4). The substantially aligned series resonance frequency (fs,SER) of the series resonators BSER and the parallel resonance frequency (fp,SH) of the shunt resonators BSH fall within the passband. FIGS. 6A through 6E provide circuit equivalents for the five phases of the response of the ladder network 40. During the first phase (phase 1, FIGS. 5C, 6A), the ladder network 40 functions to attenuate the input signal. As the series resonance frequency (fs,SH) of the shunt resonators BSH is approached, the impedance of the shunt resonators BSH drops precipitously such that the shunt resonators BSH essentially provide a short to ground at the series resonance frequency (fs,SH) of the shunt resonators (phase 2, FIGS. 5C, 6B). At the series resonance frequency (fs,SH) of the shunt resonators BSH (phase 2), the input signal is essentially blocked from the output of the ladder network 40.
Between the series resonance frequency (fs,SH) of the shunt resonators BSH and the parallel resonance frequency (fp,SER) of the series resonators BSER, which corresponds to the passband, the input signal is passed to the output with relatively little or no attenuation (phase 3, FIGS. 5C, 6C). Within the passband, the series resonators BSER present relatively low impedance, whereas the shunt resonators BSH present a relatively high impedance, wherein the combination of the two leads to a flat passband with steep low- and high-side skirts. As the parallel resonance frequency (fp,SER) of the series resonators BSER is approached, the impedance of the series resonators BSER becomes very high, such that the series resonators BSER essentially present themselves as open at the parallel resonance frequency (fp,SER) of the series resonators (phase 4, FIGS. 5C, 6D). At the parallel resonance frequency (fp,SER) of the series resonators BSER (phase 4), the input signal is again essentially blocked from the output of the ladder network 40. During the final phase (phase 5, FIGS. 5C, 6E), the ladder network 40 functions to attenuate the input signal, in a similar fashion to that provided in phase 1. As the parallel resonance frequency (fp,SER) of the series resonators BSER is passed, the impedance of the series resonators BSER decreases and the impedance of the shunt resonators BSH normalizes. Thus, the ladder network 40 functions to provide a high Q passband between the series resonance frequency (fs,SH) of the shunt resonators BSH and the parallel resonance frequency (fp,SER) of the series resonators BSER. The ladder network 40 provides extremely high attenuation at both the series resonance frequency (fs,SH) of the shunt resonators BSH and the parallel resonance frequency (fp,SER) of the series resonators. The ladder network 40 provides good attenuation below the series resonance frequency (fs,SH) of the shunt resonators BSH and above the parallel resonance frequency (fp,SER) of the series resonators BSER. As noted previously, there is a constant need to improve the performance of acoustic resonators and filters that are based thereon.